The Feeling of Power Reflection
This post is a reflection assignment from CS 480 where I read Isaac Asimov's The Feeling of Power and respond to several prompts about the story and my experience with LLMs in this course.
Asimov's The Feeling of Power: Two Questions
Q1: What does this story make me think about my own reliance on technology in daily life?
I relate to this story quite a bit. I grew up in two very different educational environments, in India and in the US. Much of my early education was in India, where calculators are not allowed in class until you reach the higher college level. Back then, I could do mental math very quickly. Since coming to the US, where calculators are widely allowed, I have relied on them so much that I have lost some of that mental math ability. It makes me feel like I have lost one of my little “superpowers,” even if it seems minor. Now, I sometimes challenge myself to solve math problems without a calculator, hoping to keep whatever skills I still have and reconnect with that sense of capability. It also makes me wonder what other skills I have lost because of my reliance of technology.
Q2: Just because a task can be automated, does that mean it should be? What are the potential consequences of relying too much on machines to do work for us?
I think we should only automate a task if one of two conditions is met. First, we already know how to complete the task and just need the outcome faster. Second, we can black box the process because it is not relevant to us. I try to make sure that if I am automating a task, one of these conditions is true so I am not losing an opportunity to improve myself.
I also think there is always a risk of relying too much on machines. If they can do everything that we can do, we become dependent on them. When it comes time for us to complete the task on our own, we may not know how to get started or how to finish it. As a result, valuable knowledge and skills could be lost.
Prompt 1: How have you used LLMs in life and other courses?
I have used LLMs in a variety of situations, most often to better understand difficult concepts, complex code, and unfamiliar technical topics. I frequently use them to build a foundational understanding of a problem before diving deeper into more detailed material. For example, when learning about agentic AI, I asked the model to explain the concept as if I were a beginner with no prior knowledge. As my understanding improved, I gradually increased the level of complexity by requesting explanations appropriate for a college student and then for someone who had completed an introductory machine learning course.
In addition to conceptual learning, I use LLMs to summarize research papers so I can quickly evaluate whether a topic aligns with my academic interests or project goals. I also use them to debug code, clarify error messages, and explore alternative approaches when solving programming assignments.
When practicing LeetCode problems, I first attempt to solve the problem on my own. After reaching a solution, I ask the model whether there is a faster or more efficient approach and request an explanation of how that solution works and why it improves performance. This helps me strengthen my problem solving skills, better understand algorithmic tradeoffs, and learn new optimization strategies rather than relying on the model for direct answers.
Server Setup & Blog Deployment Reflection
Problems You Encountered
In addition to hosting the blog on DigitalOcean using Nginx, I also wanted to set up a custom domain. This was quite a challenge because I had never done this before, and linking everything together was tricky.
How AI Was Helpful
AI was really helpful in providing a step-by-step guide to follow in order to set everything up. Whenever I ran into trouble while following the instructions, I could give it the issue or error message, and it would reevaluate the steps to fix the problem and continue guiding me until everything worked.
How AI Was Unhelpful
I would not really say there was a time when it did not help me. For every problem I encountered, I was able to use AI to troubleshoot and find a solution.
How You Refined a Prompt
I started by giving the overarching goal and then broke it down into smaller goals that could be converted into a list of TODOs. AI then generated solutions for each step. I also specified a role, such as an expert in deploying websites using Nginx and .tech domains. Another helpful trick was providing the exact error messages if I encountered any, rather than being vague, so AI could give a precise solution.
Nogramming Assignment Proposal
I use photos from my camera roll along with additional prompts to generate a vision board. AI will help by generating some of the images based on my prompts, giving me visuals I might not have found on my own. My role will be then to assemble all the images into a cohesive design using Canva. This process will allow me to combine personal photos with AI-generated visuals, creating a vision board that is both unique and meaningful to me.
~Shree